3.烟草系统卷烟需求预测模型的探讨 王森 吴春明浙江省绍兴市烟草专卖局(公司) 浙江大学计算机学院 摘要:中国烟草行业具有行政上的垄断性和生产上的计划性,有别与其它行业。做好卷烟销量的预测,是当前烟草行业工业生产环节与商业环节协同平滑进展的前提。 本文运用了时间序列预测法,并选择了带季节指数的移动平均和最小平方法,分别对烟草系统的短、中期总体销量进行预测,同时为保证预测的可信度和选择最优的预测方法,通过计算预测标准差,设定误差控制。从算例结果比较来看,预测方法具有良好的有效性和实用性。关键词:时间序列预测法 季节指数 误差控制 需求预测Discussion about Forecasting Models of Tobacco System DemandsWang Sen Wu Chunming(.Department of Computer Science and Technology, Zhejiang University , Hangzhou 310027 ,China)(.Department of Computer Science and Technology, Zhejiang University , Hangzhou 310027 ,China)Abstract:Chinese tobacco trade system is different from other trade system,it has the feature of administration monopoly and produce plan. Forecasting the tobacco Demands is the key for the industry and trade system’s mutual benefit. This paper forecaste the short-term and medium-term tobacco demands by Time-series Analysis with Season—Index Moving Average and Minimum Square method. For obtaining the excellent method, the paper calculate a standard error and set up a error control method. The cases study shows that the methods is valid and useful.Key words:Time-series Analysis, Season—Index Moving Average, error control,Demand foreasting 1.引言 就烟草行业而言,由于行业的计划性很强,烟民的消费需求变化相对较稳定,烟草行业较少存在其它行业的激烈且无序的竞争。针对这一行业特点,我们采用时间序列预测法来做卷烟销量预测 1。时间序列预测法 2 适合于变化相对稳定的预测 4 环境,算法简单有效。 时间序列预测 2是当前烟草行业当前最简单、实用的预测总量的方法,但是它无法体现中国的农历节日如中秋、春节的消费特色,无法准确预测节日性的销量增长。为提高的...